The quest to generate the perfect repository – Griffin DAL Generator

The data layer is something really important in an application. If it’s incorrectly created it will behave like a virus and infect the entire business layer making it a lot harder to build and maintain the application. Unfortunately the symptoms will not appear until the project have been built for some time. In this article I’ll present my attempt to a cure.

If you have previously read this blog you’ll probably know that I do not use ORMs. Sure, they speed up the initial development considerably. But time saved in the beginning is lost many times when the application grows and have to be maintained. The reason is that once things start to get complex the ORM will fail on you. It will return obscure error messages, fail in the mappings or not translate the LINQ statements in a way that you expected.

The entire ORM movement was created since you, me and most other devs did not want to spend so much time doing repetitive tasks in the data layer (creating all those CRUD statements and build the mappings/conversions between the DB tables and the .NET classes). However, along that way came a lot more laziness. Now we wanted to do as little as possible in the data layer, at least stuff that had to do with the DB.

Along came HQL, LINQ and other ways of abstracting away (and simplifying) the data layer access. Suddenly we have a data layer that can be compared with ASP.NET WebForms. i.e. It tries to abstract away something and act like something it isn’t. And it does it by trying to create a simple API by doing a really complex implementation.

I got two large issues with that. The first thing is that ADO.NET is in itself really simple. Anyone can write a SQL query and fetch the result. The other reason is that when something fails you have to focus on debugging the ORM instead of just correcting a SQL query.

So when something fails you have to start debugging something complex instead of something really simple. Hands on your heart, wouldn’t you prefer correcting a SQL statement instead of a mapping or invalid LINQ query?

The solution

My alternative is a generator application. It doesn’t sound so sexy (and it isn’t). But the generated code is created using my 20 years of experience in the field as starting point.

The generated code contains everything from my data access articles and is either generated using ADO.NET only or my own data mapper library (still just plain SQL). Thus, there is not a complex library which will obfuscate things or take away the control from you. Thus it’s trivial to adapt the code and to correct problems in the future.

The generates can either create Queries (as in Command/Queries) or repositories. The repositories are either generated as asynchronous or synchronous repositories, while the queries are always asynchronous.





It’s considered beta until 2015-04-30 and is free during this period (100% discount). After that the personal license will remain free while companies have to pay.

More info at the homeopage. I’ll try to implement all feature/change requests that comes in this week.

Sample repository using Griffin.Framework

Here is small sample repository using Griffin.Framework

using Griffin.Container;
using Griffin.Data.Mapper;
using OneTrueError.GlobalCore.Invitations;
using OneTrueError.GlobalCore.Invitations.Data;

namespace OneTrueError.GlobalCore.SqlServer
    internal class InvitationRepository : IInvitationRepository
        private readonly GlobalUnitOfWork _unitOfWork;

        public InvitationRepository(GlobalUnitOfWork unitOfWork)
            _unitOfWork = unitOfWork;

        public Invitation FindByEmail(string email)
            return _unitOfWork.FirstOrDefault<Invitation>(new {EmailToInvitedUser = email});

        public void Create(Invitation invitation)

        public void Update(Invitation invitation)

As you might have noticed I’m using one Assembly/DLL per SQL engine that we support. It’s not required for Griffin.Framework, but we are using plain ADO.NET for some DB operations and Griffin.Framework for others. Thus having one DLL per SQL engine allows us to write optimized queries without compromises.

In v2 of OneTrueError we are using two different databases. One for global activities like billing and customer accounts and then one customer specific DB for each customer. The reason is that we’ve had customers which gets several GB of error reports each month. Thus to make sure that the DB is responsive for all other customers (even when we are lacking a few efficient indexes) we’ve taken this approach.

When OneTrueError is started we simply use Assembly.Load together with a configuration setting in <appConfig> to be able to load the correct DLLs.

Introducing the data mapper in Griffin.Framework

As you might know I’m running a .NET exception service called OneTrueError. When I moved from a NoSQL db to SQL Azure I had to be able to work with the database in some way. I’m not using OR/Ms any more. They might significantly reduce the bootstrapping, but in the long run they always tend to make you struggle as the application grow. To me, ORMs is a bit like ASP.NET WebForms, but for data. i.e. it tries to make something what it isn’t. I therefore wanted something that did not take away the control from me nor hide the underlying layer. I still want to work with my domain entities though.

Continue reading “Introducing the data mapper in Griffin.Framework”

Repository pattern, done right

The repository pattern has been discussed a lot lately. Especially about it’s usefulness since the introduction of OR/M libraries. This post (which is the third in a series about the data layer) aims to explain why it’s still a great choice.

Continue reading “Repository pattern, done right”

ADO.NET, the right way

ADO.NET is actually quite powerful if you use it correctly. This post will teach you everything from making your ADO.NET code driver independent to how to implement the repository pattern and unit of work. This is the follow up post of my “Datalayer, the right way” post. The purpose is to demonstrate that ADO.NET can be used as an alternative to OR/Ms.

Continue reading “ADO.NET, the right way”